didrocks | good morning | 04:50 |
---|---|---|
duflu | Morning didrocks | 04:52 |
duflu | Huh? The only working call to vaCreateSurfaces is the one where we pass the parameters in the wrong order (!?)... where working means status=0 but still corruption | 04:52 |
duflu | Oh... macro magic to rearrange parameters | 04:54 |
didrocks | hey duflu, good Monday? Sounds some wayland fights ;) | 04:59 |
duflu | didrocks, It is Monday at least. Yes, Wayland :) | 05:00 |
didrocks | heh :) | 05:00 |
duflu | \o/ Hardware decoding under Wayland to Clutter! | 05:35 |
duflu | (with many hacks that now need eliminating) | 05:35 |
duflu | didrocks, Yes it is now a good Monday :) | 06:40 |
duflu | How are you? | 06:40 |
didrocks | duflu: I'm good, thanks! What happened so that it became a good Monday? :p | 06:41 |
duflu | didrocks, I finally got hardware video decoding on Wayland fixed. Only took a week | 06:41 |
didrocks | oh, that will be a great enhancement :) | 06:42 |
andyrock | good morning! | 07:00 |
didrocks | hey andyrock, good week-end? | 07:01 |
duflu | Morning andyrock | 07:01 |
andyrock | hey didrocks | 07:01 |
andyrock | hey duflu | 07:01 |
andyrock | good enough | 07:01 |
andyrock | spend the we in Bologna with my brother | 07:01 |
didrocks | sounds great ;) | 07:02 |
seb128 | good morning desktopers! | 07:07 |
seb128 | hey andyrock duflu, re didrocks | 07:09 |
duflu | Hi seb128 | 07:09 |
andyrock | hey seb128 | 07:09 |
didrocks | re seb128 | 07:09 |
Laney | what's up | 08:03 |
willcooke | morning all! | 08:04 |
willcooke | Doing a pretty good impression of winter here today | 08:04 |
seb128 | hey Laney, willcooke, how is u.k today? | 08:08 |
seb128 | more automn like here | 08:09 |
=== maclin1 is now known as maclin | ||
Laney | hey seb128 | 08:11 |
Laney | yeah quite grey, wet and windy | 08:11 |
flexiondotorg | Morning seb128 Laney willcooke andyrock duflu | 08:12 |
duflu | Hi flexiondotorg | 08:12 |
seb128 | hey flexiondotorg | 08:13 |
alexarnaud | good morning all :) ! | 08:16 |
seb128 | hey alexarnaud | 08:17 |
alexarnaud | seb128: how are you today? | 08:17 |
seb128 | good! you? | 08:19 |
=== JamieBen_ is now known as JamieBennett | ||
alexarnaud | seb128: good :) | 09:21 |
=== carlolo is now known as clobrano | ||
infinity | seb128: Who do I complain to that since rebooting from unity to gnome-shell, I can no longer disable my touchpad? | 10:41 |
didrocks | infinity: could be the -synaptics vs libinput transition? | 10:42 |
infinity | I don't think my driver changed. | 10:42 |
infinity | But the mouse/keyboard options sure did. | 10:42 |
seb128 | infinity, uninstall synaptic if it's installed | 10:42 |
didrocks | bug #1686081 | 10:42 |
ubot5 | bug 1686081 in xorg (Ubuntu) "If -synaptics is installed, GNOME Mouse & Touchpad Settings doesn't work" [High,Confirmed] https://launchpad.net/bugs/1686081 | 10:42 |
infinity | Hrm. Will try and report back. | 10:43 |
seb128 | seems didrocks is on the case, /me goes back to what he was doing | 10:43 |
infinity | didrocks: You win. How very unintuitive. | 10:47 |
infinity | didrocks: But thanks for saving me from going insane with accidental palm drags. | 10:47 |
didrocks | infinity: let's say "have experience with it" :-) | 10:48 |
didrocks | happy that it helped! | 10:48 |
willcooke | didrocks, do we have a "proper" solution to that, like can we forcibly remove synaptics or something? | 10:49 |
infinity | didrocks: I assume there's still ongoing work to make the default theme less... Fedora? | 10:50 |
infinity | (All the blue is rather jarringly unUbuntu) | 10:50 |
willcooke | infinity, yeah we do - got a hackfest kinda thing planned for end of August | 10:50 |
willcooke | infinity, I made a start. kinda. http://imgur.com/a/NjUXa | 10:51 |
willcooke | few things to fix there ;) | 10:51 |
infinity | willcooke: Cool. I think in the interim, only a few colours need changing to make it at least kinda look right (the +/- bars in the panel, and the default lock screen colour) | 10:51 |
didrocks | willcooke: there is a trello card for it: https://trello.com/c/9GI3EFh2/122-bug1686081-if-synaptics-is-installed-gnome-mouse-touchpad-settings-doesnt-work, discussed about it with seb128, not trivial, but really annoying | 10:51 |
infinity | willcooke: Oh, and maybe the top bar slightly less absolute black. :P | 10:51 |
willcooke | infinity, yeah totally agree. Thats what I was trying to fix actually. global search and replace was a bad idea it seems. who knew?! | 10:52 |
infinity | (ie: same background as our gnome-terminal) | 10:52 |
willcooke | didrocks, thanks! | 10:52 |
didrocks | infinity: I guess we all are in favor of blackless top bar (at least, with our theme :p) | 10:52 |
seb128 | willcooke, it's the same old "need to port u-c-c to libinput" that oem wants and we are talking about for some cycles | 10:52 |
infinity | didrocks: I actually like the default GNOME theme, I just don't want it identical to Fedora. And we don't use true black anywhere, so that same "purplish almost-black" that we use in the terminal background would be appropriate. | 10:53 |
=== Kamilion is now known as |{amilion | ||
=== |{amilion is now known as Kamilion | ||
didrocks | infinity: could be, we'll experiment (and maybe provide 2 sessions: gnome upstream one with default themes et our own "ubuntu experience"), that's still to be discussed :) | 10:54 |
infinity | And, of course, something orangeish to replace the blue +/- bars. | 10:54 |
infinity | I'm going to miss menus in my window titlenbars. | 10:54 |
infinity | titlebars, too. | 10:54 |
didrocks | (same) | 10:55 |
infinity | As horrible hacks go, when it finally worked how I wanted it, it was niceish. | 10:55 |
didrocks | what have you changed? | 10:55 |
didrocks | just curious, could be useful feedbacks | 10:55 |
didrocks | any extensions/anything? | 10:55 |
infinity | Basically nothing. Literally just logged into the fresh session a few hours ago. | 10:55 |
infinity | I assume it migrated over some generic settings from unity (ie: it's still got my wallpaper selection), but for the most part, it looks very Fedora. | 10:56 |
didrocks | apart from the launcher and some icon renames, nothing is ported, the keys were still the generic one | 10:56 |
didrocks | so nothing to port was the net benefit :) | 10:56 |
infinity | Right, I didn't mean to imply there was actual migration, just that it's obvious some stuff went unchanged, like who renders the root window (nautilus?) | 10:57 |
infinity | Oddly enough, that's sort of comforting. Even when your whole workflow goes to crap, having the wallpaper consistent feels like not all is lost. :P | 10:58 |
didrocks | ahah :) | 10:58 |
didrocks | that familiar feeling… | 10:58 |
infinity | didrocks: Anyhow, I won't be shy about providing some feedback as I find things that rub me the wrong way. | 11:02 |
willcooke | thanks infinity, all feedback gratefully received | 11:02 |
infinity | didrocks: First thing I noted (other than the synaptics bug) was that Super-L wasn't bound to lock screen anymore. That must have been a Unity special. | 11:02 |
didrocks | I'm sure you won't be shy ;) and yeah, please feel free :) | 11:02 |
willcooke | haha! Told you so seb128 & Laney ;D ^ | 11:03 |
infinity | didrocks: But given that binding also works in Windows, I found it a pleasant analog. | 11:03 |
didrocks | yeah, their default is the old ctrl+alt+l? | 11:03 |
infinity | *nod* | 11:03 |
infinity | I think both worked for me in Unity because I got Super+L from Unity and Ctrl-Alt-L from the gnome keyboard settings. | 11:03 |
infinity | On relog, of course, I only had the latter. | 11:03 |
infinity | And since they don't seem allow binding two things to one action, I now have the former. :P | 11:04 |
didrocks | yeah, I never left Ctrl+Alt+l, the other one was added to compiz as a secondary action | 11:04 |
infinity | I use a Windows desktop for gaming, and Super-L is the shortcut there, and I dig consistency. | 11:04 |
infinity | But also, it's the binding that makes more sense when Super is "they key to manage overall desktop stuff". | 11:05 |
didrocks | willcooke: can you add it to your special list of polish that we can discuss about? | 11:05 |
infinity | (Though, I guess maybe GNOME people still live in a pretend world where some people's keyboards were made prior to 1997?) | 11:05 |
infinity | Or Happy Hackers. | 11:05 |
didrocks | but worskpace changes are still Ctrl+Alt + something | 11:05 |
infinity | Yeah. True. But also no analog there on other OSes, which may be why I care less. ;) | 11:06 |
didrocks | I remeber we had the plan for super + anything on unity, we only did half of it though | 11:06 |
didrocks | ;) | 11:06 |
willcooke | didrocks, I'll create a trello card to track | 11:06 |
didrocks | great ;) | 11:06 |
seb128 | I think we should have both keybindings to lock working | 11:06 |
infinity | ^ | 11:06 |
infinity | I'd agree with that. | 11:06 |
seb128 | if we pick one we are going to piss half of the users either way | 11:06 |
infinity | Is there a way under the hood to make GNOME take two bindings? | 11:06 |
infinity | The UI sure won't let you. | 11:06 |
seb128 | I don't think so | 11:06 |
seb128 | which is the issue | 11:06 |
infinity | Ick. | 11:06 |
seb128 | we might need to hack and special case ctrl-alt-L or something | 11:07 |
infinity | I guess you could just add a second binding "Alternate Desktop Lock". | 11:07 |
willcooke | I think I found an old patch to move the gsetting to an array | 11:07 |
infinity | Which calls the same thing. | 11:07 |
seb128 | or that | 11:07 |
seb128 | willcooke, right, that's more work/incompatible changes though | 11:07 |
didrocks | willcooke: the issue isn't doing it, it's more about exposing it with minimal changes in g-c-c | 11:07 |
willcooke | ack | 11:07 |
infinity | I mean, I think it would be stellar to allow an array for ANY binding, but a second binding would be a reasonable hack for this one small regression. | 11:07 |
infinity | And simple. | 11:08 |
didrocks | yeah, but as seb128 told, it means we need to change all gesttings keys from string to list of array | 11:08 |
didrocks | and patch every apps which are using them | 11:08 |
infinity | Right, that's a mess. | 11:08 |
infinity | Hence the other thing. | 11:08 |
didrocks | doesn't seem minimal, not going to work in the snap world sharing the same keys from other code | 11:08 |
didrocks | yeah | 11:08 |
seb128 | but yeah, having "lock screen" and "alternative lock screen" lines wouldn't be the end of hte world | 11:08 |
didrocks | I guess making that one special is fine as well | 11:08 |
seb128 | the other way would be to special case one in the shell | 11:08 |
seb128 | which is what we did in unity it looks like | 11:09 |
seb128 | and we never got a complain about it | 11:09 |
infinity | Special casing one makes it effectively invisible to the user. | 11:09 |
infinity | But fair play, since that's how it was before too. :P | 11:09 |
seb128 | right | 11:09 |
seb128 | and you could say that the documented one is the one in the settings | 11:09 |
infinity | (Though it did show up in the Unity help screen, I think, so only "invisible" fromthe POV of the keyboard bindings screen) | 11:09 |
seb128 | and the other one is the compat one of people who are too used to the old keybinding | 11:09 |
infinity | seb128: I think one of my favourite things about having a Unity7 and Windows7 (now Win10) desktop side-by-side was that all the important bindings were effectively identical. | 11:11 |
infinity | Super+Num to pick from the launcher, Super+L to lock, etc. | 11:12 |
infinity | I suspect that wasn't an accident. | 11:12 |
infinity | Really, the only noticeable difference between them is that one of them can play more video games, and WINDOWS UPDATE IS STILL A FLAMING HEAP OF OH GOD HOW HAVE THEY NOT IMPROVED THIS IN TWENTY YEARS ARGH. | 11:13 |
infinity | Otherwise, they're hard to tell apart. | 11:13 |
infinity | (Seriously, the next time you're hating on the performance of dpkg and apt, go update a fresh Windows installation and get some perspective) | 11:14 |
infinity | Oh nice, Firefox seems to have survived the transition. It only shows me a menu bar when I press Alt, like the upstream builds do. | 11:20 |
infinity | (PS: I took your survery, please don't switch the default browser, kthx) | 11:21 |
seb128 | infinity, we wouldn't have to deal with builds on weird archs if chromium was our default :p | 11:23 |
infinity | I'm pretty sure I've had to fix Chromium on armhf in the past. | 11:23 |
seb128 | right, armhf is something we need to deal with | 11:24 |
seb128 | ppc64el or s390x not so much | 11:25 |
infinity | If we get no commitment from IBM on those, we'll drop them. | 11:25 |
infinity | I don't object to dropping things with zero upstream support, I object to the knee-jerk "it didn't build, so let's not look into it and discuss dropping it". | 11:26 |
infinity | Since there have been many times in the past when it was a simple 1-liner. | 11:26 |
infinity | Well, s390x is already dropped, so that's a red herring people keep bringing up. | 11:27 |
seb128 | it's not "it didn't build", it's "firefox is outdated for most of the cycle and our unstable users have a version with security issues" | 11:27 |
infinity | But I'll talk to the POWER guys about ppc64el. | 11:27 |
seb128 | and that has been the case for at least 3 cycles now | 11:27 |
seb128 | it's not a one time thing | 11:28 |
seb128 | it has been pretty much a permanent situation since xenial | 11:28 |
infinity | It's not a 1-time thing, but it's also different arches each time. | 11:28 |
ricotz | jfyi, current failures of firefox 55 https://launchpad.net/%7Emozillateam/+archive/ubuntu/firefox-next/+sourcepub/8091927/+listing-archive-extra | 11:28 |
seb128 | so arm64 in addition of what artful currently has | 11:29 |
infinity | In xenial, it's currently armhf and ppc64el. In trusty, it's arm64 and ppc64el. Which is curious. | 11:29 |
ricotz | ff 56 for comparsion https://launchpad.net/~ubuntu-mozilla-daily/+archive/ubuntu/ppa/+sourcepub/8099928/+listing-archive-extra | 11:29 |
infinity | Given identical upstream codebases, if armhf fails on one series and arm64 on another, it might be our bug, not upstream's. | 11:30 |
ricotz | (arm64 is happy there) | 11:30 |
ricotz | (some failures are rust related) | 11:31 |
seb128 | 56 is missing cargo on ppc64el | 11:31 |
ricotz | seb128, I know, I hacked some packages myself | 11:32 |
ricotz | I think chrisccoulson is onto updating those officially | 11:32 |
ricotz | https://bugs.launchpad.net/ubuntu/+source/cargo/+bug/1701556 | 11:33 |
ubot5 | Ubuntu bug 1701556 in rustc (Ubuntu) "Backport rustc 1.17 and cargo 0.18 to 14.04/16.04 and 17.04" [Undecided,New] | 11:33 |
infinity | didrocks: So, plugins. Is there anything (you know of) that will restore my ability to see several timezones in my clock applet? | 11:35 |
* infinity decides that maybe trying to sleep is smarter than worrying about what timezone he's in. | 11:38 | |
infinity | Oh, I like the virtual desktop upper bound just being "one more than you're using" instead of a static number. | 11:38 |
seb128 | infinity, you can add locations from gnome-clocks and they should be listed | 11:39 |
seb128 | (we currently don't pre-install gnome-clocks though) | 11:39 |
jbicha | good morning | 11:39 |
seb128 | hey jbicha | 11:39 |
infinity | seb128: Ahh. I was going to say "That's not a command I have". :P | 11:39 |
infinity | seb128: Hrm. I have them in gnome-clocks, but that doesn't change the applet drop-down in any meaningful way. | 11:41 |
jbicha | I proposed a gjs SRU … which resulted in this email: https://lists.ubuntu.com/archives/ubuntu-release/2017-July/004163.html | 11:41 |
infinity | jbicha: Robie's very much a letter of the law guy. Which isn't the worst thing ever (I'd rather that than the opposite), but yeah, probably means we need more law for him to follow. | 11:42 |
infinity | jbicha: FTR, if he'd passed on it, you could have poked me directly. | 11:43 |
seb128 | infinity, right, it isn't working for me either but jbicha said that was supposed to do that iirc, I didn't have time to debug/have a look yet | 11:43 |
jbicha | the current status is that Robie added the GNOME microrelease exception to the SRU wiki page | 11:44 |
infinity | GNOME's always had an MRE, literally since we first had SRUs. | 11:44 |
infinity | But we may have undocumented it when we went to the new blanket statements. | 11:44 |
infinity | Oops. | 11:44 |
jbicha | yes, it was removed from the wiki but it's back now | 11:45 |
jbicha | so assuming there are no objections, it should be a bit smoother for GNOME SRUs now :) | 11:45 |
seb128 | infinity, in fact it works today for me, maybe it needed a session restart after installing gnome-clocks or something... it's listed in the popup you get when clicking on the date in the top panel | 11:47 |
infinity | seb128: Ahh, lemme log out, I guess. | 11:47 |
infinity | seb128: Well, one bug down, one new one discovered. | 11:49 |
infinity | seb128: Seems on a fresh boot, Ctrl-Alt-T gives me terminals in a snappy and quick timeframe. When I log out and back in (WTF?), Ctrl-Alt-T has a delay of some 10-20 seconds. | 11:50 |
infinity | seb128: Thought this was a 1-time weirdness last time I did a log out dance, but it just reproduced this time too. | 11:50 |
seb128 | dunno about this one | 11:50 |
infinity | seb128: Have you seen that, or am I special/insane? | 11:51 |
infinity | I guess I'm special. | 11:51 |
seb128 | I didn't see it | 11:51 |
infinity | seb128: Must be something long-running between sessions, but I kinda assumed logins would be under a systemd user session and the cgroup blown away on logout (maybe I'm wrong there) | 11:52 |
infinity | In fact, all keyboard shortcuts have that massive delay. Lock screen was the same. | 11:54 |
infinity | Weeeeeird. | 11:54 |
infinity | I wonder if it's some conflict between the two gsd-keyboard sessions running. Though, they should both be running on a fresh boot/login too. | 11:55 |
jbicha | if there was a cgroup that was killed on logout, maybe we wouldn't have had LP: #1610944 | 11:55 |
ubot5 | Launchpad bug 1610944 in gnome-session (Ubuntu Zesty) "GNOME Online Accounts breaks if you log out (until you reboot)" [High,Triaged] https://launchpad.net/bugs/1610944 | 11:55 |
jbicha | upstream has a different workaround as of last week but not in artful yet https://bugzilla.gnome.org/764029 | 11:55 |
ubot5 | Gnome bug 764029 in general "goa-daemon (and most other D-Bus services) not stopped when the user session goes away" [Critical,New] | 11:55 |
infinity | jbicha: Ahh, hrm. Well, that is probably related to the bug I'm seeing now. | 11:56 |
jbicha | is one of the gsd-keyboard's from gdm3? because that would make sense since gdm3 itself runs a minimal gnome-shell session | 11:57 |
infinity | I find it strange that the distro that foisted systemd on us isn't using it correctly. | 11:57 |
infinity | jbicha: Yeah, one's gdm, and one's me. Like I said, I'm sure they're always both running, so that's a red herring. | 11:58 |
infinity | jbicha: But clearly, some state is breaking on log out/in. | 11:58 |
jbicha | that GOA bug was a release blocker for much of Fedora 26's release cycle | 11:58 |
infinity | jbicha: And it's odd breakage, since I'd expect shortcut keys to either work or not work, rather than working after a 15s delay. :P | 11:58 |
ogra_ | just add a splash screen :P | 12:00 |
infinity | A nice modal dialog that says "GNOME has detected you used a keybinding shortcut, we'll process your request and act on it shortly..." | 12:01 |
ogra_ | :) | 12:01 |
infinity | Oh well, for now I'll reboot. That one will surely drive me batty in short order and I'll file a bug and/or debug it myself. | 12:02 |
infinity | But yes, it would almost certainly "just go away" if we used a proper user session. | 12:02 |
infinity | Murdering all the things is a simple fix for so many bugs. | 12:02 |
jbicha | willcooke: wow https://imgur.com/a/NjUXa is very orange | 13:03 |
willcooke | jbicha, ha! I got a bit carried away | 13:03 |
ogra_ | we'll just ship sunglasses along with the isos | 13:08 |
willcooke | lol | 13:09 |
seb128 | Laney, https://trello.com/c/9kiXF8rW/134-systemd-user-session-for-gnome-shell ... do we have no systemd user session at all in GNOME atm? how did it start under unity/would it be difficult to do the same for the GNOME session? | 13:15 |
seb128 | I don't remember the details | 13:15 |
Laney | seb128: Exec=/usr/lib/gnome-session/run-systemd-session unity-session.target | 13:16 |
seb128 | so we "just" need an equivalent target under GNOME I guess? | 13:17 |
Laney | dunno | 13:18 |
Laney | that would be the first thing to try | 13:18 |
Laney | I'm not sure if there would be issues or not | 13:18 |
seb128 | do we feel like we need that for 17.10? | 13:18 |
seb128 | I wonder if we have things that got made systemd user jobs that are not going to work anymore without that | 13:19 |
Laney | that was all downstream stuff, so I doubt it | 13:19 |
seb128 | hum, k | 13:22 |
seb128 | going to keep investigating why file sharing over bluetooth is not working then | 13:22 |
seb128 | it looked like it could be due to that but maybe not | 13:22 |
seb128 | thanks Laney | 13:22 |
Laney | there could be issues like things not being killed when you log out | 13:23 |
seb128 | Laney, fedora has http://pkgs.fedoraproject.org/cgit/rpms/bluez.git/tree/0001-Allow-using-obexd-without-systemd-in-the-user-sessio.patch | 13:24 |
seb128 | Laney, I don't think logout is the issue there, I think bluez just expect the user session to be there | 13:25 |
seb128 | I guess we could use the fedora workaround | 13:25 |
* seb128 tries that | 13:25 | |
Laney | seb128: that SystemdService specified unit doesn't exist | 13:29 |
Laney | Failed to introspect object / of service org.bluez.obex: Unit dbus-org.bluez.obex.service not found. | 13:30 |
Laney | https://paste.ubuntu.com/25162502/ | 13:31 |
* mancman3 waves @ the good non ms winshit users | 13:32 | |
seb128 | Laney, you are right | 13:36 |
Laney | seb128: bluetooth.service has Alias=dbus-org.bluez.service | 13:37 |
seb128 | Laney, typo you think? | 13:38 |
Laney | maybe? | 13:38 |
Laney | not sure what it's supposed to be | 13:39 |
Laney | but maybe it's a missing alias on obex.service? | 13:39 |
Laney | no, that is there | 13:40 |
Laney | I think the obex package should enable the unit (or create that symlink) | 13:40 |
seb128 | hum, ctrl-R at the wrong place | 13:41 |
seb128 | Laney, I'm a bit confused now, isn't it simply the SystemdService being wrong as your sed-ed earlier? | 13:42 |
seb128 | or is that not enough? | 13:42 |
Laney | seb128: the value it had should work because obex.service has an alias for it | 13:42 |
Laney | but it doesn't because the unit isn't enabled by the package | 13:43 |
Laney | you can fix it by making a symlink | 13:43 |
seb128 | Laney, thanks | 13:45 |
seb128 | ricotz, hey, did you get any reply from meson upstream about the armhf issue? | 13:50 |
ricotz | seb128, no, I guess since doko proposed the other arm fixes, he pretty sure knows how to fix the remaining ones | 13:52 |
seb128 | ricotz, did you ask him about it? | 13:52 |
ricotz | seb128, no | 13:53 |
seb128 | ricotz, could you? ;-) | 13:53 |
ricotz | seb128, you are afraid? ;) | 13:53 |
seb128 | no but you are the one that asked for that version to be synced and you seem to understand the issue | 13:54 |
seb128 | Laney, sorry but I'm going to bother you again about that systemd thing, the right way to enable it would be to create a symlink to the service in a <service-used-by-default>.wants? | 14:18 |
seb128 | Laney, is there a standard "service-used-by-default" we use for such cases? I tried to look for examples and failed to find some | 14:18 |
seb128 | well https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=764678 has sockets.target.want but it's for a socket unit | 14:19 |
ubot5 | Debian bug 764678 in debhelper "dh-systemd: Please support systemd user services" [Wishlist,Open] | 14:19 |
seb128 | didrocks, ^ or maybe you know? | 14:19 |
Laney | seb128: just in /usr/lib/systemd/user/ | 14:20 |
didrocks | depends, user services or system service? | 14:20 |
seb128 | didrocks, systemd --user | 14:21 |
didrocks | system services should use dh_systemd to enable it, indeed | 14:21 |
Laney | it doesn't need to be in wants of anything | 14:21 |
didrocks | ah | 14:21 |
seb128 | how do you call that? | 14:21 |
Laney | this is dbus activation | 14:21 |
didrocks | I don't know if dh_systemd has the support for it, I would say no | 14:21 |
seb128 | didrocks, cf the bug I just mentioned | 14:21 |
seb128 | it doesn't | 14:21 |
seb128 | that's why I'm asking about the symlink ;-) | 14:21 |
ricotz | seb128, regarding meson, basically https://paste.debian.net/plain/977929 | 14:21 |
didrocks | seb128: symlink is fine, it's basically what dh_systemd does (+ service start/restart) if you are in the systemd case | 14:22 |
seb128 | ricotz, can you send that upstream? (and maybe a debdiff to Debian or us for packaging) | 14:22 |
didrocks | it doesn't use systemctl enable, because of systemd bootstrapping issue | 14:22 |
didrocks | (like, when you get to your first release with systemd, but it's not installed yet, so you don't have systemctl, and so on…) | 14:22 |
seb128 | hum | 14:23 |
seb128 | I'm getting confused | 14:23 |
didrocks | if it's an user service, a symlink to the corresponding .wants to enable it is enough | 14:24 |
seb128 | Laney, there is a obex.service already in that dir but you said the issue is that it's not enabled and would need a symlink? | 14:24 |
Laney | yes | 14:24 |
Laney | but not to wants | 14:24 |
seb128 | to where? | 14:24 |
didrocks | ah, I see, Laney says it's dbus activated | 14:24 |
seb128 | right | 14:24 |
Laney | the value specified in alias= | 14:24 |
Laney | that's what systemctl --user enable foo would do | 14:24 |
Laney | you have to do that manually, that's what didrocks is saying | 14:24 |
seb128 | why do we need an alias at all here? | 14:25 |
seb128 | things seems more complicated that they should be | 14:25 |
Laney | because the dbus service file is asking for that unit to be started | 14:25 |
seb128 | why don't we just the dbus service start the right unit? | 14:25 |
didrocks | (I'm having the same, probably stupid question than seb) | 14:25 |
Laney | they expect the unit to be enabled | 14:26 |
Laney | that is a fair expectation | 14:26 |
seb128 | what unit? | 14:26 |
didrocks | there are multiple provider? | 14:26 |
didrocks | (hence the alias?) | 14:26 |
Laney | don't know | 14:26 |
Laney | but it is fair enough for something to rely on a unit being enabled | 14:26 |
seb128 | I guess I don't understand why we need an alias | 14:26 |
seb128 | rather than having 1 unit | 14:26 |
seb128 | and the dbus activation used that 1 unit | 14:26 |
seb128 | using | 14:26 |
Laney | no idea, it wasn't any of us that developed this | 14:27 |
Laney | I'm telling you how to make it work with how it is set up now | 14:27 |
seb128 | k, so you are not suggesting the design is better | 14:27 |
didrocks | Laney: where is dbus activation file btw which reference that alias? (just curious) | 14:27 |
seb128 | just hitting on what to do with what upstream is providing? | 14:27 |
Laney | /usr/share/dbus-1/services/org.bluez.obex.service | 14:27 |
seb128 | Laney, gotcha, sorry for being slow to get it | 14:27 |
Laney | I have no idea if that design is optimal or not | 14:27 |
seb128 | Laney, thanks again! | 14:27 |
Laney | presumably there is a reason they did it this way | 14:27 |
Laney | np! | 14:27 |
didrocks | maybe they want a non version alias where maybe there will be versioned one | 14:27 |
ricotz | seb128, https://paste.debian.net/plain/977932 | 14:27 |
didrocks | or multiple prociders | 14:27 |
didrocks | providers* | 14:28 |
seb128 | jbicha, do you think you could get that patch from Rico in debian ^? | 14:28 |
seb128 | didrocks, it's a bit difficult to get proper context on bluez bugs, they don't have proper bug tracking&such | 14:29 |
didrocks | do you know how this dbus activation is working, as there is no service, it seems as you get the bug it doesn't fallback to Exec= | 14:29 |
didrocks | correct? | 14:29 |
Laney | that's right | 14:29 |
ricotz | seb128, jbicha, note this an attempt | 14:30 |
didrocks | ok, so there is no point in keeping the Exec= as soon as you have SystemdService= | 14:30 |
didrocks | oh, or maybe dbus is smart enough to say "no systemd running, I'm ignoring SystemdService and only using Exec=" | 14:30 |
Laney | yep | 14:31 |
seb128 | didrocks, http://pkgs.fedoraproject.org/cgit/rpms/bluez.git/tree/0001-Allow-using-obexd-without-systemd-in-the-user-sessio.patch suggests it does | 14:31 |
didrocks | interesting, didn't follow those user sessions changes | 14:31 |
seb128 | (the comment/description) | 14:31 |
Laney | it's for if you don't have systemd --user | 14:31 |
didrocks | yeah, making sense | 14:31 |
Laney | yeah you can see that the old file in that patch was buggy in that case | 14:31 |
didrocks | oh ok, so yeah, disabling it unconditionnally ofc is wrong | 14:32 |
didrocks | thanks for the quick tech catchup Laney, seb128 ;) | 14:32 |
seb128 | thanks didrocks Laney ;-) | 14:32 |
Laney | #ubuntu-desktop Tech Talks | 14:32 |
didrocks | :-) | 14:33 |
Laney | /m/me needs a tech talk | 14:33 |
Laney | ffs | 14:33 |
Laney | and better wifi | 14:33 |
didrocks | not a better keyboard? :) | 14:34 |
Laney | or mosh or something | 14:34 |
didrocks | heh | 14:34 |
Laney | trying to work with GMainContext, threads and such | 14:35 |
Laney | it's confusing | 14:35 |
didrocks | another threading bug your are fighting against? | 14:36 |
Laney | same | 14:36 |
didrocks | argh, good luck :( | 14:36 |
Laney | sendto(3<socket:[2178330]>, "GET /v2/system-info HTTP/1.1\r\nHost: \r\nConnection: keep-alive\r\nUser-Agent: snapd-glib/1.16\r\n\r\n", 93, MSG_NOSIGNAL, NULL, 0) = 93 | 14:37 |
Laney | poll([{fd=3<socket:[2178330]>, events=POLLIN}, {fd=6<anon_inode:[eventfd]>, events=POLLIN}], 2, -1 | 14:37 |
Laney | why does that poll block? | 14:37 |
Laney | there should be shit coming back from snapd | 14:37 |
didrocks | I guess you can't mock easily the daemon-side of the API because there are other exchanges first? | 14:38 |
Laney | nah | 14:38 |
Laney | I made a minimal version that does the same thing | 14:38 |
Laney | works ofc | 14:38 |
didrocks | so, snapd doesn't answer? | 14:38 |
didrocks | or sounds like it doesn't | 14:39 |
jbicha | seb128: the Debian maintainer of meson == meson upstream maintainer | 14:39 |
Laney | pretty sure it does | 14:39 |
didrocks | hum | 14:39 |
Laney | but for some reason I don't get it | 14:39 |
seb128 | jbicha, but he refuses to get fixes in Debian? | 14:39 |
Laney | it's all GSocket and stuff at the application level of course | 14:39 |
* Laney shrugs | 14:39 | |
Laney | back in a min | 14:40 |
seb128 | jbicha, the issue is that the meson updates are blocked in artful-proposed due to armhf autopkgtest issues, I asked rico if he could ping upstream about it and he came with that patch but I don't think he upstreamed it | 14:41 |
jbicha | seb128: I don't think he's really refusing, it's just the patches need to be proposed and maybe pushed a bit | 14:41 |
ricotz | seb128, https://github.com/mesonbuild/meson/pull/2110 | 14:41 |
seb128 | ricotz, thanks | 14:41 |
seb128 | jbicha, seems ricotz did upstream it now, so unping | 14:41 |
jbicha | :) | 14:42 |
ricotz | as said I haven't/can't confirmed that it actually fixes it | 14:42 |
seb128 | it's fine | 14:42 |
seb128 | now that it's upstreamed maybe Jussi looks at it | 14:42 |
jbicha | ricotz: if you push it to a PPA, I can run an autopkgtest on it to test the fix | 14:45 |
jbicha | because autopkgtests can be run from a PPA but it needs to be run by someone with upload rights | 14:46 |
ricotz | jbicha, ok, should appear here https://launchpad.net/~ricotz/+archive/ubuntu/staging/+packages?field.name_filter=&field.status_filter=published&field.series_filter=artful | 14:47 |
ricotz | seb128, that reminds me, what happened with poppler? | 14:48 |
seb128 | ricotz, I said I would wait for libreoffice to migrate to not makie it part of another transition | 14:49 |
ricotz | seb128, ok | 14:49 |
seb128 | ricotz, or libreoffice is having failing autopkgtests now so it's not migrating (maybe it's also blocked by other things) | 14:49 |
jbicha | should we ignore the LO autopkgtests on i386 too (and maybe s390x?)? | 14:50 |
ricotz | jbicha, interesting so it would be possible to autopkgtests on libreoffice of ppa:libreoffice/libreoffice-prereleases ? | 14:50 |
seb128 | jbicha, why? | 14:51 |
ricotz | java-stack kernel bug ;) | 14:51 |
jbicha | seb128: for the same reason we ignored build tests on i386 :( | 14:51 |
jbicha | ricotz: yes but neither you nor Sweetshark had upload rights… | 14:52 |
seb128 | jbicha, is the failure due to the i386/kernel/java segfault? | 14:52 |
ricotz | jbicha, ok | 14:53 |
seb128 | jbicha, s390x is green after my retry from earlier | 14:53 |
seb128 | not on -l10n though | 14:53 |
seb128 | but yeah, +1 to skip on i386 | 14:53 |
seb128 | who can do that? L_aney? | 14:53 |
jbicha | oh, now LO is stuck because of python-defaults | 14:54 |
seb128 | bah | 14:55 |
ricotz | jbicha, the meson package is built | 15:43 |
jbicha | ricotz: autopkgtest succeeds on amd64 and armhf, only 2 arches I tried | 15:53 |
jbicha | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful-ricotz-staging/artful/armhf/m/meson/20170724_155058_dbc59@/log.gz | 15:53 |
jbicha | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful-ricotz-staging/artful/amd64/m/meson/20170724_154005_dbc59@/log.gz | 15:53 |
jbicha | we can push that update to artful but I'd prefer to wait for Jussi's review since I don't think it's very urgent | 15:55 |
ricotz | jbicha, ok | 15:56 |
ricotz | jbicha, it got merged | 16:18 |
jbicha | yes, could you ask him whether he'll do a Debian upload for it or if we should just upload to Ubuntu directly if we want the fix sooner? | 16:20 |
=== JanC is now known as Guest39739 | ||
=== JanC_ is now known as JanC | ||
Laney | /run-snapctl/basic: OK | 16:36 |
Laney | laney@raleigh (master↑1|✚3)> echo $? ~/dev/canonical/upstream/random/snapd-glib | 16:36 |
Laney | 0 | 16:36 |
* Laney dies | 16:36 | |
Laney | the testsuite's mock was relying on the library using main contexts in a particular way that I had changed | 16:41 |
ricotz | jbicha, since you are in the meson channel you can read the response | 16:51 |
ricotz | seb128, jbicha, so basically if you want meson to be fixed asap, just push my package | 16:58 |
=== Guest49031 is now known as fredp | ||
=== fredp is now known as Guest83180 | ||
ricotz | jbicha, could you run the autopkgtest for this amd64 build https://launchpad.net/~libreoffice/+archive/ubuntu/libreoffice-prereleases/+build/13139327 ? | 18:51 |
jbicha | done | 18:57 |
jbicha | to find the results for the earlier test you can visit | 18:58 |
jbicha | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful-ricotz-staging | 18:58 |
jbicha | look for the <name> field with the log you want to see, like | 18:59 |
jbicha | artful/armhf/m/meson/20170724_155058_dbc59@/log.gz | 18:59 |
jbicha | then add that to the end of the original URL | 18:59 |
jbicha | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful-ricotz-staging/artful/armhf/m/meson/20170724_155058_dbc59@/log.gz | 18:59 |
jbicha | it won't be live until the test finishes, but I believe the URL should be | 19:00 |
jbicha | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful-libreoffice-prereleases | 19:00 |
ricotz | jbicha, thanks! | 19:02 |
jbicha | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful-libreoffice-libreoffice-prereleases | 19:03 |
jbicha | better URL ^ | 19:03 |
jbicha | ok, trying again with all-proposed because of the python3 transition fun | 19:05 |
jbicha | my 2nd try probably will fail too, it was my 3rd try that used all-proposed | 19:06 |
ricotz | jbicha, http://autopkgtest.ubuntu.com/running#pkg-libreoffice | 19:08 |
willcooke | night all | 20:41 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!